Better, Faster
Decisions with Data.
The Context Factory trusted by frontier teams.
From the team behind Uber’s Michelangelo AI platform.
LATAM Airlines · Customer story
Delphina makes it easy for everyone
to finally make decisions based on data.
DelphinaWhy is retention falling this month?
NRR fell to 98% this month, from 104%. Most of the drop is SMB downgrades after the March price change.
✓ Net revenue retention · approved
Slack · #finance@Delphina Why did gross margin dip in Q3?
Q3 gross margin was 71.2%, down 1.8 pts. Cloud costs on two enterprise migrations explain most of it.
✓ Gross margin · approved
ClaudeAre we on track for the Q4 pipeline target?
Not yet. Qualified pipeline is at 82% of target with six weeks left; EMEA is the gap at 61%.
✓ Qualified pipeline · approved
ChatGPTWhat is CAC payback on paid social?
Paid social CAC payback is 14 months, up from 11 in Q2. Rising Meta CPMs drive most of it.
✓ CAC payback · approved
Context FactoryContext in use+ 11 more files
- Metric
- Business context
- Glossary
✓ Approved by Finance
- Governed definitions & metrics
- Curated by AI, approved by SMEs
- Self-healing as things change12 fixes this week
Delphina maps metrics and business context, keeps SMEs in the loop and automatically updates as things change. Available in any AI frontdoor.
Loved by CEOs who want more from their data.
“Delphina gives us AI superpowers for data. It’s changed how we make decisions.”
Chris Best
CEO
“10 years in this business, and I’m still finding new insights every day through Delphina.”
Waleed Nasr
CEO
“I stopped waiting for reports and started finding answers myself.”
Zach Gordon
Co-CEO
“We’re an airline that thinks of ourselves as a tech company. Delphina is a core part of how we’re delivering on that.”
Andrés Bucchi
CDO & Chief Engineering Officer
Flywheel agents create & maintain the context, so you can trust the answers.
analytics.delphina.ai / context-factory
Context from your sources
Read from dbt, dashboards, docs, and query logs, and kept current as they change.
Evals done right
Tests from trusted dashboards leverage what your team has already verified.
Fixes land in your knowledge base
Changes land versioned in a knowledge base that you own.
Agents find and fix gaps,
so accuracy compounds.
Six steps. Agents run five.
One turn · about a minute
reactivated customers by region
AnswerSteps
I don’t have a definition for reactivated customer yet, and I’d rather not guess. I’ve opened an investigation to find who owns it — I’ll follow up here with an answer you can trust.
KB-418 · Define “reactivated customer” Investigating
✓ Investigation finished · 8 min · definition confirmed by Priya Nair
4,812 reactivated customers in Q2, using the definition Priya confirmed: orders again after 90+ days without an order, free trials excluded. NA leads at 2,140; APAC is the fastest-growing region.
Reactivated Customers by Region — Q2 2026
Source: fct_orders · definition v1 (Priya Nair)
Ask Delphina...
Context Factory · Issues
OpenAgent-owned
KB-418Open
Define “reactivated customer”
- Search dbt, Looker, Notion · no definition found
- Expert graph → Priya Nair, owner of fct_orders
- Ask Priya in #data-help→ Slack · #data-help · 9:46 am@Priya quick one from a question Maya asked in Delphina: how do we define a reactivated customer? Two mentions in Notion, no definition, and you own fct_orders.
- Priya answers · fact approved✓ Published to knowledge base · 9:52 amReactivated customer — a customer who places an order after 90+ days without one. Free-trial accounts excluded.v1 · Priya Nair · approved in Slack
- Re-run Maya’s question · answer changed
- Save as a regression test
- Close · notify Maya
# data-help
14
DelphinaAPP9:46 AM
Hi @Priya — quick one from a question Maya asked in Delphina. How do we define a reactivated customer? I found two mentions in Notion but no definition, and you own
fct_orders.P
Priya Nair9:51 AM
DelphinaAPP9:51 AM
Thanks. Here’s what I’ll add to the knowledge base:
Reactivated customer — a customer who places an order after 90+ days without one. Free-trial accounts excluded.Source: Priya Nair (Growth) · supersedes: none
✓ Approved · 9:52 AM
DelphinaAPP9:53 AM
Added, and Maya’s answer is regenerated: 4,812 reactivated customers in Q2. Saved as a test so it can’t silently regress. Thanks — that was about 20 seconds of your time.
Fast to deploy. Faster to trust and scale.
Fastest to…
Before
With Delphina
Build context
One day.
Baseline context distilled from what you already have.
BeforeTwo quarters of modeling before anyone trusts an answer.
Build context
Two quarters of modeling before anyone trusts an answer.
One day.
Baseline context distilled from what you already have.
Absorb what's new
Same day.
Delphina never stops re-reading your sources.
BeforeNew tables and use cases wait for the next sprint.
Absorb what's new
New tables and use cases wait for the next sprint.
Same day.
Delphina never stops re-reading your sources.
Fix a stale definition
30 seconds.
Approved once, fixed for everyone, forever.
BeforeA Jira ticket, fixed next quarter. Maybe.
Fix a stale definition
A Jira ticket, fixed next quarter. Maybe.
30 seconds.
Approved once, fixed for everyone, forever.
Ask and build
Seconds to minutes.
Answers, research reports, and data apps.
BeforeThe analyst queue for an answer. Next quarter's OKRs for a new dashboard.
Ask and build
The analyst queue for an answer. Next quarter's OKRs for a new dashboard.
Seconds to minutes.
Answers, research reports, and data apps.
Scale outside the data team
Same day.
Delphina only answers when it knows. Across our customers, 73% of active users are outside the data team.
BeforeNever, for most teams. The context is always a quarter behind and quietly going stale.
Scale outside the data team
Never, for most teams. The context is always a quarter behind and quietly going stale.
Same day.
Delphina only answers when it knows. Across our customers, 73% of active users are outside the data team.
Semantic layers never realized their potential.
Semantic layer projects lag and decay.
- They lag. Model first. Learn what matters later.
- They decay. The business moves. The definitions don’t.
- AI writes the wrong thing faster. Still a guess. Still decays.
The factory learns in minutes from real demand.
- Start from evidence. Query logs, dashboards, and dbt models.
- Follow demand. Every question shows what matters now.
- Agents do the work. Draft, test, ship.
Ship apps that take action.
Build an app that pulls from business tools, exposes interactive analysis on top, and pushes the results out to the systems your team runs on.
Pulls live from
Snowflake
Pipeline & LTV
Salesforce
Opportunities
Google Sheets
Q4 budget plan
analytics.delphina.ai / apps / budget-planner
Live · 2 min ago
Q4 paid media planner
Approved definitions| Channel | Q4 spend | CAC | Pipeline / $ |
|---|---|---|---|
| Google Search | $40.0K | $820 | 6.2× |
| Meta | $60.0K | $1,450 | 2.9× |
| $25.0K | $1,900 | 2.1× |
Projected pipeline
+$0K
Blended CAC
0.0%
Ask this app anything…
Updates budgets in Google Ads, Meta & LinkedIn
Pushes to
Google Ads
$40.0K
Search budget
Meta Ads
$60.0K
Campaign budget
LinkedIn Ads
$25.0K
Campaign budget
Enterprise-grade security is built in from the foundation.
AI on your laptop with data + internet access is risky
- 1The agent uses its API key to pull company data.
- 2It reads a web page with a prompt injection.
- 3It sends your data to an unknown server outside your company.
Vulnerable to exfiltration via prompt injection. Ask Claude about it.
With Delphina
-
No Open Internet AccessCode runs in an isolated Firecracker microVM.
-
Every Message is LoggedEvery prompt, query, & tool call is logged and exportable.
-
Hosted in our Secure CloudAll raw data is processed in our AWS cloud or your VPC.
SOC 2
Type II Certified
Annual audit. Continuous controls monitoring.
Access
Read-only, least-privilege
Delphina connects with per-user read-only credentials.
Tenancy
Single-tenant data plane
Every customer gets a separate, isolated data plane.
Sandbox
Isolated execution
Python and DuckDB run in an isolated Firecracker microVM with no internet access.
Audit
Structured logs
Every prompt, query, and knowledge reference logged and exportable.
Your data can do 1,000× more than it could last year.
Will you capture it, or let it slip?
Meet Delphina.
The Context Factory trusted by frontier teams.